My autoscaled instance fails when running "git clone" on a private repo.
I do have the SSH key placed at /root/.ssh/id_rsa
on the machine, and when I SSH into the machine and run sudo su; git clone <the repo>
it succeeds.
Also, in the extra_vm_bash_script
field: I added a whoami
command which prints root
, so it seems like the user being used to run the git clone during task execution is in fact root
.
For context, here's the startup command that the autoscaler runs:
python -m clearml_agent --config-file /root/clearml.conf daemon --queue aws_4gpu_machines --docker python:3.9
Full log included...
